Failover stability?

David W. Hankins David_Hankins at isc.org
Wed Jul 5 16:28:35 UTC 2006


On Tue, Jul 04, 2006 at 02:05:13PM -0400, Keith Lawson wrote:
> I'm just wondering what the state is of master/slave peer failover. Is
> failover stable enough to be used in a production environment now? 
> 
> I've been testing using 3.0.4 one one subnet  and I'm running into a
> scenario where the master load balances a request to the slave. The
> slave then ignores the request and logs a "peer holds all free leases"
> message. I've been reading through the mailing list archives and
> indications are that these problems are supposed to be resolved in 3.0.4
> but that doesn't seem to be the case. 

Something's going wrong with your setup.  That log line tells me
that your secondary thinks all the leases are in the 'free' state.

Are you sure your servers are operating in 'normal' failover state?

Once they connect for the first time, the primary should give away
half its free leases to the secondary (moving them to 'backup' state).


Make sure both your servers are using precisely the same configuration
file (modulo failover {} config bits which have to be different).

The easiest way to do that is to make dhcpd.conf contain only the
failover config, and to move the rest of the config to a separate
file ("dhcpd.common.conf", say), then use an include directive:

	include "dhcpd.common.conf";

Then just copy this file between servers to keep it consistent.

-- 
David W. Hankins		"If you don't do it right the first time,
Software Engineer			you'll just have to do it again."
Internet Systems Consortium, Inc.		-- Jack T. Hankins


More information about the dhcp-users mailing list